Azure Meetup-Budget tank of 91 IoT

The premise of my Azure Meetup presentation was could you build an interesting project on a rainy weekend afternoon with a constrained budget (tank of 91 octane petrol) and minimal soldering .

Budget

Our family car is a VW Passat V6 4Motion which has a 62 Litre tank. The driver usually doesn’t usually stop to fill up until the fuel light has been on for a bit which helped.

PetrolReceipt

Based on the most recent receipt the budget was NZD132.

Where possible I purchased parts locally (the tech equivalent of food miles) or on special.

My bill of materials (prices as at 2018-06) was on budget.

The devDuino V2.2 and nRF24L01 module were USD26.20 approx. NZD37.50 (including freight) from elecrow.

Tradeoffs

I powered my Raspberry PI with a spare cellphone charger (make sure it can supply enough current to reliably power the device).

The devDuino V2.has an ATSHA204A which provides a guaranteed unique 72-bit serial number (makes it harder to screw up provisioning devices in the field).

I use a 32G MicroSD rather than a 16G MicroSD card as I have had issued with 16G cards getting corrupted by more recent upgrades (possibly running out of space?)

The Raspberry PI shield requires a simple modification to enable interrupt driven operation.

My sample devDuino V2.2 client uses an external temperature and humidity sensor, modifying this code to use the onboard temperature sensor an MCP9700 will be covered in another post.

The devDuino V2 is a little bit cheaper USD15.99 NZD37.31, has the same onboard temperature sensor as the V2.2 but no unique serial number chip.

The devDuino V4.0 has an onboard HTU21D temperature + humidity sensor but no unique serial number and the batteries are expensive.

The code and deployment instructions for the nRF24L01 field gateway applications for AdaFruit.IO and Azure IoT Hub/Azure IoT Central are available on hackster.IO.

RPiWithnRF24Plate

AdaFruit.IO has free and USD10.00/month options which work well for many hobbyist projects.

AdaFruitIO

Wireless field gateway protocol V1

I’m going to build a number of nRF2L01P field gateways (Netduino Ethernet & Wifi running .NetMF, Raspberry PI running Windows 10 IoT Core, RedBearLab 3200  etc.), clients which run on a variety of hardware (Arduino, devDuino, Netduino, Seeeduino etc.) which, then upload data to a selection of IoT Cloud services (AdaFruit.IO, ThingSpeak, Microsoft IoT Central etc.)

The nRF24L01P is widely supported with messages up to 32 bytes long, low power consumption and 250kbps, 1Mbps and 2Mbps data rates.

The aim is to keep the protocol simple (telemetry only initially) to implement and debug as the client side code will be utilised by high school student projects.

The first byte of the message specifies the message type

0 = Echo

The message is displayed by the field gateway as text & hexadecimal.

1 = Device identifier + Comma separated values (CSV) payload

[0] – Set to 1

[1] – Device identifier length

[2]..[2+Device identifier length] – Unique device identifier bytes e.g. Mac address

[2+Device identifier length+1 ]..[31] – CSV payload e.g.  SensorID value, SensorID value

Overtime I will support more message types and wireless protocols.

 

nRF24 Windows 10 IoT Core Background Task

First step is to build a basic Windows 10 IoT Core background task which can receive and display messages sent from a variety of devices across an nRF24L01 wireless link.

If you create a new “Windows IoT Core” “Background Application” project then copy this code into StartupTasks.cs the namespace has to be changed in the C# file, project properties\library\Default namespace and “Package.appxmanifest”\declarations\Entry Point.

/*

Copyright ® 2017 December devMobile Software, All Rights Reserved

THIS CODE AND INFORMATION IS PROVIDED "AS IS" WITHOUT WARRANTY OF ANY
KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE
IMPLIED WARRANTIES OF MERCHANTABILITY AND/OR FITNESS FOR A PARTICULAR
PURPOSE.

http://www.devmobile.co.nz

*/
using System;
using System.Diagnostics;
using System.Text;
using Radios.RF24;
using Windows.ApplicationModel.Background;

namespace devmobile.IoTCore.nRF24BackgroundTask
{
    public sealed class StartupTask : IBackgroundTask
    {
      private const byte ChipEnablePin = 25;
      private const byte ChipSelectPin = 0;
      private const byte nRF24InterruptPin = 17;
      private const string BaseStationAddress = "Base1";
      private const byte nRF24Channel = 10;
      private RF24 Radio = new RF24();
      private BackgroundTaskDeferral deferral;

      public void Run(IBackgroundTaskInstance taskInstance)
        {
         Radio.OnDataReceived += Radio_OnDataReceived;
         Radio.OnTransmitFailed += Radio_OnTransmitFailed;
         Radio.OnTransmitSuccess += Radio_OnTransmitSuccess;

         Radio.Initialize(ChipEnablePin, ChipSelectPin, nRF24InterruptPin);
         Radio.Address = Encoding.UTF8.GetBytes(BaseStationAddress);
         Radio.Channel = nRF24Channel;
         Radio.PowerLevel = PowerLevel.High;
         Radio.DataRate = DataRate.DR250Kbps;
         Radio.IsEnabled = true;

         Debug.WriteLine("Address: " + Encoding.UTF8.GetString(Radio.Address));
         Debug.WriteLine("PA: " + Radio.PowerLevel);
         Debug.WriteLine("IsAutoAcknowledge: " + Radio.IsAutoAcknowledge);
         Debug.WriteLine("Channel: " + Radio.Channel);
         Debug.WriteLine("DataRate: " + Radio.DataRate);
         Debug.WriteLine("IsDynamicAcknowledge: " + Radio.IsDyanmicAcknowledge);
         Debug.WriteLine("IsDynamicPayload: " + Radio.IsDynamicPayload);
         Debug.WriteLine("IsEnabled: " + Radio.IsEnabled);
         Debug.WriteLine("Frequency: " + Radio.Frequency);
         Debug.WriteLine("IsInitialized: " + Radio.IsInitialized);
         Debug.WriteLine("IsPowered: " + Radio.IsPowered);

         deferral = taskInstance.GetDeferral();

         Debug.WriteLine("Run completed");
      }

      private void Radio_OnDataReceived(byte[] data)
      {
         // Display as Unicode
         string unicodeText = Encoding.UTF8.GetString(data);
         Debug.WriteLine("Unicode - Payload Length {0} Unicode Length {1} Unicode text {2}", data.Length, unicodeText.Length, unicodeText);

         // display as hex
         Debug.WriteLine("Hex - Length {0} Payload {1}", data.Length, BitConverter.ToString(data));
      }

      private void Radio_OnTransmitSuccess()
      {
         Debug.WriteLine("Transmit Succeeded!");
      }

      private void Radio_OnTransmitFailed()
      {
         Debug.WriteLine("Transmit Failed!");
      }
   }
}

This was displayed in the output window of Visual Studio

Address: Base1
PA: 15
IsAutoAcknowledge: True
Channel: 10
DataRate: DR250Kbps
IsDynamicAcknowledge: False
IsDynamicPayload: True
IsEnabled: True
Frequency: 2410
IsInitialized: True
IsPowered: True
Run completed

Interrupt Triggered: FallingEdge
Unicode – Payload Length 19 Unicode Length 19 Unicode text T  23.8,H  73,V 3.26
Hex – Length 19 Payload 54-20-32-33-2E-38-2C-48-20-20-37-33-2C-56-20-33-2E-32-36
Interrupt Triggered: RisingEdge

Note the odd formatting of the Temperature and humidity values which is due to the way dtostrf function in the Atmel AVR library works.

Also noticed the techfooninja nRF24 library has configurable output power level which I will try to retrofit onto the Gralin NetMF library.

Next, several simple Arduino, devDuino V2.2, Seeeduino V4.2 and Netduino 2/3 clients (plus possibly some others)

nRF24 Windows 10 IoT Core reboot

My first live deployment of the nRF24L01 Windows 10 IoT Core field gateway is now scheduled for mid Q1 2018 so time for a reboot. After digging out my Raspbery PI 2/3 devices and the nRF24L01+ shield (with modifications detailed here) I have a basic plan with some milestones.

My aim is to be able to wirelessly acquire data from several dozen Arduino, devduino, seeeduino, and Netduino devices, Then, using a field gateway on a Raspberry PI running Windows 10 IoT Core upload it to Microsoft IoT Central

First bit of code – Bleepy a simple background application to test the piezo beeper on the RPI NRF24 Shield

namespace devmobile.IoTCore.Bleepy
{
   public sealed class StartupTask : IBackgroundTask
   {
      private BackgroundTaskDeferral deferral;
      private const int ledPinNumber = 4;
      private GpioPin ledGpioPin;
      private ThreadPoolTimer timer;

      public void Run(IBackgroundTaskInstance taskInstance)
      {
         var gpioController = GpioController.GetDefault();
         if (gpioController == null)
         {
            Debug.WriteLine("GpioController.GetDefault failed");
            return;
         }

         ledGpioPin = gpioController.OpenPin(ledPinNumber);
         if (ledGpioPin == null)
         {
            Debug.WriteLine("gpioController.OpenPin failed");
            return;
         }

         ledGpioPin.SetDriveMode(GpioPinDriveMode.Output);

         this.timer = ThreadPoolTimer.CreatePeriodicTimer(Timer_Tick, TimeSpan.FromMilliseconds(500));

         deferral = taskInstance.GetDeferral();

         Debug.WriteLine("Rum completed");
      }

      private void Timer_Tick(ThreadPoolTimer timer)
      {
         GpioPinValue currentPinValue = ledGpioPin.Read();

         if (currentPinValue == GpioPinValue.High)
         {
            ledGpioPin.Write(GpioPinValue.Low);
         }
         else
         {
            ledGpioPin.Write(GpioPinValue.High);
         }
      }
   }
}

Note the blob of blu tack over the piezo beeper to mute noise
nRF24ShieldMuted

nRF24 Windows 10 IoT Core Hardware

Taking my own advice I decided to purchase a couple of Raspberry Pi to NRF24L01 shields from Ceech a vendor on Tindie.

The nRF24L01 libraries for my .Net Micro framework and WIndows 10 IoT Core devices use an interrupt driver approach rather than polling status registers to see what is going on.

Like most Raspberry PI shields intended to be used with a *nix based operating system the interrupt pin was not connected to a General Purpose Input/Output (GPIO) pin.

NRF24PiPlateModification

My first step was to add a jumper wire from the pin 8 on the nRF24L01 to GPIO pin 17 on Raspberry PI connector.

I then downloaded the techfooninja Radios.RF24 library for Windows IoT core and update the configuration to suit my modifcations. In the TestApp the modifications were limited to changing the interrupt pin from GPI 4 to GPO 17

private const byte IRQ_PIN = 4;

private const byte IRQ_PIN = 17;

I used a socket for the nRF24L01 device so I can trial different devices, for a production system I would solder the device to the shield to improve reliability.

RPiWithnRF24Plate

I then ran the my test application software in a stress test rig overnight to check for any reliability issues. The 5 x netduino devices were sending messages every 500mSec

RPIStressTester

MS Ignite Auckland NZ Presentation now available online

My presentation “All your device are belong to us” [M240] is now online at MSDN Channel 9

So much hype, so many different devices, so many protocols, so much data, so little security, welcome to the Internet of Things. Come and see how you can build an affordable, securable, scalable, interoperable, robust & reliable solution with embedded devices, Windows 10 IoT and Microsoft Azure. By 2020 there will be 26 Billion devices and 4.5 million developers building solutions so the scope is limitless.

I had 8 devices in my presentation so the scope for disaster was high.

The first demo was of how sensors could be connected across Arduino, Netduino and Raspberry PI platforms.

The Arduino demo used

The Netduino demo used

The Raspbery PI Windows 10 IoT Core demo used

The hobbyist data acquisition demo collected data from two devduino devices that were in passed around by the audience and were each equipped with a Temperature & Humidity sensor. They uploaded data to Xively over an NRF24L01 link to a gateway running on a Netduino 3 Ethernet and the data was displayed in real-time on my house information page

The professional data acquisition demo uploaded telemetry data to an Azure ServiceBus EventHub and retrieved commands from an Azure ServiceBus Queue. Both devices were running software based on Azure ServiceBus Lite by Paolo Paiterno

The telemetry stream was the temperature of some iced water.

The commands were processed by a Raspbery PI running Windows 10 IoT Core which turned a small fan on & off to illustrate how a FrostFan could be used in a vineyard to reduce frost damage to the vines.

Frost Fan demo

MS Ignite 2015 Frost Fan demo

My demos all worked on the day which was a major win as many other presenters struggled with connectivity. Thanks to the conference infrastructure support guys who helped me sort things out.

With the benefit of hindsight, I tried to fit too much in and the overnight partial rewrite post attending the presentation Mashup the Internet of Things, Azure App Service and Windows 10 to Deliver Business Value [M387] by Rob Tiffany was a bit rushed.